poi读取word中的目录大纲,导入

在做一个导入word的业务时,发现poi并不能很好的读取到文档中的目录和级别,这段代码做个记录

cpp 复制代码
 public ResponseResult importDirectory(MultipartFile file){
        try {
            InputStream inputStream = file.getInputStream();
            XWPFDocument xdoc = new XWPFDocument(inputStream);
            List<XWPFParagraph> paragraphs = xdoc.getParagraphs();
            List<ReadDto> readDtos = new ArrayList<>();
            for (XWPFParagraph paragraph : paragraphs) {
                String text = paragraph.getText();
                String titleLvl = getTitleLvl(xdoc, paragraph);
                if (StringUtil.isNotEmpty(titleLvl)) {
                    int level = Integer.valueOf(titleLvl);
//                System.out.println("text: " + text + ", titleLvl: " + titleLvl);
                    ReadDto readDto = new ReadDto();
                    readDto.setText(text);
                    readDto.setTitleLevel(level);
                    readDtos.add(readDto);
                }
            }
            int zeroCount = 0;//0出现的次数
            int oneCount = 0;//1出现的次数
            int twoCount = 0;//2出现的次数
            int threeCount = 0;//3出现的次数
            int curPoint = 0;//当前指针值
            for (int i = 0; i < readDtos.size(); i++) {
                int curLevel = readDtos.get(i).getTitleLevel();
                if (curLevel > 4) {
                    throw new BusinessException("暂不支持目录层级超过4层!!!");
                }
                if (curPoint == 0) {
                    zeroCount++;
                    curPoint = 1;
                    readDtos.get(i).setOne(zeroCount);
                } else if (curPoint == 1) {
                    if (curLevel == 0) {
                        zeroCount++;
                        oneCount = 0;
                        twoCount = 0;
                        threeCount = 0;
                        curPoint = 1;
                        readDtos.get(i).setOne(zeroCount);
                    }
                    if (curLevel == 1) {
                        curPoint++;
                        oneCount++;
                        readDtos.get(i).setOne(zeroCount);
                        readDtos.get(i).setTwo(oneCount);
                    }
                } else if (curPoint == 2) {
                    if (curLevel == 0) {
                        zeroCount++;
                        oneCount = 0;
                        twoCount = 0;
                        threeCount = 0;
                        curPoint = 1;
                        readDtos.get(i).setOne(zeroCount);
                    } else if (curLevel == 1) {
                        oneCount++;
                        twoCount = 0;
                        curPoint = 2;
                        readDtos.get(i).setOne(zeroCount);
                        readDtos.get(i).setTwo(oneCount);
                    } else if (curLevel == 2) {
                        curPoint = 3;
                        twoCount++;
                        threeCount = 0;
                        readDtos.get(i).setOne(zeroCount);
                        readDtos.get(i).setTwo(oneCount);
                        readDtos.get(i).setThr(twoCount);
                    }
                } else if (curPoint == 3) {
                    if (curLevel == 0) {
                        zeroCount++;
                        oneCount = 0;
                        twoCount = 0;
                        threeCount = 0;
                        curPoint = 1;
                        readDtos.get(i).setOne(zeroCount);
                    } else if (curLevel == 1) {
                        oneCount++;
                        curPoint = 2;
                        twoCount = 0;
                        readDtos.get(i).setOne(zeroCount);
                        readDtos.get(i).setTwo(oneCount);
                    } else if (curLevel == 2) {
                        curPoint = 3;
                        twoCount++;
                        threeCount = 0;
                        readDtos.get(i).setOne(zeroCount);
                        readDtos.get(i).setTwo(oneCount);
                        readDtos.get(i).setThr(twoCount);
                    } else if (curLevel == 3) {
                        threeCount++;
                        if (i < readDtos.size() - 1) {
                            int nextLevel = readDtos.get(i + 1).getTitleLevel();
                            if (nextLevel > 3) {
                                throw new BusinessException("暂不支持目录层级超过4层!!!");
                            }
                            if (nextLevel == 3) {
                                curPoint = 3;
                            } else if (nextLevel < 3) {
                                curPoint = nextLevel + 1;
                            }
                        }
                        readDtos.get(i).setOne(zeroCount);
                        readDtos.get(i).setTwo(oneCount);
                        readDtos.get(i).setThr(twoCount);
                        readDtos.get(i).setFou(threeCount);
                    }
                }
            }
            System.out.println(readDtos);//读取数据结果集合
            return status ? ResponseResult.success("操作成功") : ResponseResult.failure("操作失败");

        } catch (IOException e) {
            e.printStackTrace();
            throw new RuntimeException(e);
        }
//        return ResponseResult.failure("操作失败");
    }

    private static String getTitleLvl(XWPFDocument doc, XWPFParagraph para) {
        String titleLvl = "";
        try {
            //判断该段落是否设置了大纲级别
            if (para.getCTP().getPPr().getOutlineLvl() != null) {
                return String.valueOf(para.getCTP().getPPr().getOutlineLvl().getVal());
            }
        } catch (Exception e) {
        }
        try {
            //判断该段落的样式是否设置了大纲级别
            if (doc.getStyles().getStyle(para.getStyle()).getCTStyle().getPPr().getOutlineLvl() != null) {
                return String.valueOf(doc.getStyles().getStyle(para.getStyle()).getCTStyle().getPPr().getOutlineLvl().getVal());
            }
        } catch (Exception e) {
        }
        try {
            //判断该段落的样式的基础样式是否设置了大纲级别
            if (doc.getStyles().getStyle(doc.getStyles().getStyle(para.getStyle()).getCTStyle().getBasedOn().getVal())
                    .getCTStyle().getPPr().getOutlineLvl() != null) {
                String styleName = doc.getStyles().getStyle(para.getStyle()).getCTStyle().getBasedOn().getVal();
                return String.valueOf(doc.getStyles().getStyle(styleName).getCTStyle().getPPr().getOutlineLvl().getVal());
            }
        } catch (Exception e) {

        }
        try {
            if (para.getStyleID() != null) {
                return para.getStyleID();
            }
        } catch (Exception e) {

        }

        return titleLvl;
    }
相关推荐
yivifu12 小时前
使用VBA区分简体中文段落和繁体中文段落的方法
word·excel·vba
骆驼爱记录1 天前
Word三线表制作全攻略
自动化·word·wps·新人首发
盘古工具2 天前
告别重复劳动:Word 重复上一次操作的两种方法
windows·word
好名字08213 天前
Vue2转Word方法(html-docx-js库)
javascript·html·word
檐下翻书1733 天前
企业组织架构图导出Word 在线编辑免费工具
人工智能·信息可视化·去中心化·word·流程图·ai编程
涛涛北京3 天前
【word】- 宏操作 - 如何在word中插入宏
word
涛涛北京3 天前
【word】- 宏操作 - 如何在word中批量修改图片格式
word
usdoc文档预览3 天前
Word文件转PDF、WPS在线打印、js提取Word文件内容、轻松将Word文档转为PDF
pdf·word·wps·在线预览·word转pdf·在线编辑·ofd在线预览解析
身如柳絮随风扬3 天前
Spring Boot文档导出(Word,Excel,PPT),导入(Excel)
spring boot·word·excel·ppt
赵丙双3 天前
python-docx 报错 KeyError: “There is no item named ‘NULL‘ in the archive“
python·word·docx·python-docx